Towards a Domestication Strategy for Indigenous Fruit Trees in the Tropics - Indigenous Fruit Trees in the Tropics: Domestication, Utilization and Commercialization - page 36

wild or unimproved population from which new selections can be derived. The
selection population is the somewhat improved population of genotypes which
are being tested and which are used in subsequent breeding programmes to
create the next generation of potential cultivars. A wide range of genotypes may
be kept in this population as long as each has at least one characteristic of
possible future interest. The production population consists of the highly selected
genotypes, which are used for planting.
As mentioned earlier, there are two basic approaches to the genetic
improvement of trees: the seed-based breeding approach typical of forestry,
and the clonal approach typical of horticulture. The seed approach typically
involves the selection of populations (provenance testing) and/or families
(progeny testing) (Zobel and Talbert, 1984; Leakey, 1991). While this approach
could be taken for indigenous fruit trees, it is very likely that an examination of
the ten situations outlined in the Introduction would indicate that a clonal
approach is more appropriate. There are basically two ways to select the best
individual trees for cloning from broad and diverse wild populations: (i)
selection from a pool of seedlings of virtually unknown quality in a nursery or
field trial (although it may be known that the pool originates from a good
provenance or progeny); and (ii) selection of proven mature trees in wild or
planted populations (Fig. 2.2). In scenario (i), genetic improvement in yield per
hectare will undoubtedly require a series of tests, each spanning many years.
Typically, there are four levels of testing (Foster and Bertolucci, 1994):
1. Initial screening with large numbers, preferably tens of thousands, of
seedlings or, if seedlings have already been cloned, a few ramets per clone.
2. Candidacy testing with large numbers of cloned genotypes fewer than with
initial screening (preferably hundreds or thousands) and two to six ramets per
clone.
3. Clonal performance trials with moderate numbers of clones (e.g. fewer than
200) and large numbers (e.g. 0.1 ha plots) of ramets per clone.
4. Compatibility trials with small numbers of clones (e.g. 20-50) with very
large plot sizes.
It is important to recognize that there is a trade-off between the accuracy of
genetic value estimation and the intensity of selection (i.e. greater accuracy is at
the expense of numbers of families, individuals per family, or clones). For cost-
effective clonal tree improvement programmes with limited or fixed resources, it
has been found that the best strategy is to plant as many clones as possible
with relatively few ramets per clone.
To short-circuit the lengthy process of field trials, Ladipo et al . (1991a, b)
developed a predictive test for timber tree seedlings in which the initial
screening is done on young seedlings in the nursery; it is then possible to jump
straight into clonal or compatibility trials with some confidence. To date there is
no similar opportunity for fruit trees.
Like the predictive test, scenario (i) is an alternative and much quicker
option. In this case, mature trees, which have already expressed their genetic
potential at a particular site over many years of growth, are selected and
propagated vegetatively and the propagules are planted either in clonal